Update app.py

app.py

@@ -1,39 +1,53 @@
 import streamlit as st
+import pandas as pd
 from transformers import pipeline, AutoTokenizer
 import base64
 
-# Load the GPT-2 text generation model from Hugging Face
-text_generator = pipeline("text-generation", model="Shruti9756/G24_Contract_Summarization_step3")
+# Load the EasyTerms/legalSummerizerET model from Hugging Face
+summarizer = pipeline("summarization", model="Shruti9756/G24_Contract_Summarization_step3")
 
 # Increase the maximum token limit
 tokenizer = AutoTokenizer.from_pretrained("Shruti9756/G24_Contract_Summarization_step3")
-text_generator.model.config.max_position_embeddings = tokenizer.model_max_length
+summarizer.model.config.max_position_embeddings = tokenizer.model_max_length
 
-# Function to generate text using the GPT-2 model
-def generate_text(prompt):
-    generated_text = text_generator(prompt, max_length=300, num_return_sequences=1, temperature=0.7)[0]['generated_text']
-    return generated_text
+# Function to generate summary using the EasyTerms/legalSummerizerET model
+def generate_summary(contract_text):
+    summary = summarizer(contract_text, max_length=512, min_length=50, length_penalty=2.0, num_beams=4, early_stopping=True)
+    return summary[0]['summary_text']
 
-# # Function to handle feedback and store it in a CSV file
-# def handle_feedback(feedback_data, feedback_file):
-#     # Your existing feedback handling logic remains the same
+# Function to handle feedback and store it in a CSV file
+def handle_feedback(feedback_data, feedback_file):
+    feedback_df = pd.DataFrame(feedback_data, columns=['Contract', 'Summary', '👍', '👎'])
 
-# # Function to create a download link for a binary file
-# def get_binary_file_downloader_html(file_path, file_label):
-#     # Your existing file downloader logic remains the same
+    # Save the dataframe to the feedback CSV file
+    feedback_df.to_csv(feedback_file, mode='a', index=False, header=not st.session_state.feedback_csv_exists)
+
+    # Display a feedback collected message only if thumbs up or thumbs down is clicked
+    if '👍' in feedback_df['👍'].values or '👎' in feedback_df['👎'].values:
+        st.success("Feedback collected successfully!")
+
+        # Display a download button for the user
+        st.markdown(get_binary_file_downloader_html(feedback_file, 'Feedback Data'), unsafe_allow_html=True)
+
+# Function to create a download link for a binary file
+def get_binary_file_downloader_html(file_path, file_label):
+    with open(file_path, 'rb') as file:
+        file_content = file.read()
+    b64 = base64.b64encode(file_content).decode()
+    return f'<a href="data:file/csv;base64,{b64}" download="{file_label}.csv">Click here to download {file_label}</a>'
 
 # Main Streamlit app
 def main():
-    st.title("Text Generation with Feedback")
+    st.title("Legal Contract Summarizer with Feedback")
 
-    # Input area for prompt
-    prompt = st.text_area("Enter a prompt:", height=100)
+    # Input area for legal contract
+    contract_text = st.text_area("Enter the legal contract:", height=200)  # Increase the height to handle larger contracts
 
-    # Button to generate text
-    if st.button("Generate Text"):
-        generated_text = generate_text(prompt)
-        st.subheader("Generated Text:")
-        st.write(generated_text)
+    # Button to generate summary
+    if st.button("Generate Summary"):
+        summary = generate_summary(contract_text)
+        st.subheader("Generated Summary:")
+        st.write(summary)
 
         # Feedback section
         st.subheader("Feedback:")
@@ -43,16 +57,16 @@ def main():
         chosen = "👍" if thumbs_up else None
         rejected = "👎" if thumbs_down else None
 
-        # feedback_data.append((prompt, generated_text, chosen, rejected))
+        feedback_data.append((contract_text, summary, chosen, rejected))
 
         # Handle feedback data
-        # if feedback_data:
-        #     feedback_file = 'feedback.csv'
-        #     st.session_state.feedback_csv_exists = True
-        #     handle_feedback(feedback_data, feedback_file)
+        if feedback_data:
+            feedback_file = 'feedback.csv'
+            st.session_state.feedback_csv_exists = True
+            handle_feedback(feedback_data, feedback_file)
 
 # Initialize feedback data
-# feedback_data = []
+feedback_data = []
 
 # Run the app
 if __name__ == "__main__":